The Unimodality of the Crank on Overpartitions

Wenston J.T. Zang1 and Helen W.J. Zhang2

arXiv:1811.10013v1 [math.CO] 25 Nov 2018

1
Institute of Advanced Study of Mathematics
Harbin Institute of Technology, Heilongjiang 150001, P.R. China
2
Center for Applied Mathematics
Tianjin University, Tianjin 300072, P.R. China
1 2
zang@hit.edu.cn, wenjingzhang@tju.edu.cn

Abstract. Let N(m, n) denote the number of partitions of n with rank m, and let
M(m, n) denote the number of partitions of n with crank m. Chan and Mao proved that
for any nonnegative integers m and n, N(m, n) ≥ N(m + 2, n) and for any nonnegative
integers m and n such that n ≥ 12, n 6= m + 2, N(m, n) ≥ N(m, n − 1). Recently, Ji
and Zang showed that for n ≥ 44 and 1 ≤ m ≤ n − 1, M(m − 1, n) ≥ M(m, n) and
for n ≥ 14 and 0 ≤ m ≤ n − 2, M(m, n) ≥ M(m, n − 1). In this paper, we analogue
the result of Ji and Zang to overpartitions. Note that Bringmann, Lovejoy and Osburn
introduced two type of cranks on overpartitions, namely the first residue crank and the
second residue crank. Consequently, for the first residue crank M (m, n), we show that
M (m − 1, n) ≥ M (m, n) for m ≥ 1 and n ≥ 3 and M (m, n) ≥ M(m, n + 1) for m ≥ 0 and
n ≥ 1. For the second residue crank M2(m, n), we show that M2(m − 1, n) ≥ M2(m, n)
for m ≥ 1 and n ≥ 0 and M2(m, n) ≥ M2(m, n + 1) for m ≥ 0 and n ≥ 1. Moreover,
let Mk (m, n) denote the number of k-colored partitions of n with k-crank m, which was
defined by Fu and Tang. They conjectured that when k ≥ 2, Mk (m − 1, n) ≥ Mk (m, n)
except for k = 2 and n = 1. With the aid of the inequality M(m − 1, n) ≥ M (m, n) for
m ≥ 1 and n ≥ 3, we confirm this conjecture.
Keywords: Overpartition, crank, unimodal, k-colored partition.

AMS Classifications: 11P81, 05A17, 05A20.

1 Introduction
The first residue crank and the second residue crank on overpartitions are two important
statistics in the theory of partitions. The main objective of this paper is to investigate
the distribution of the first residue and the second residue on overpartitions. Recall that
the rank of a partition was defined by Dyson [16] as the largest part minus the number of
parts. The crank of a partition was introduced by Andrews and Garvan [4] as the largest
part if the partition contains no ones, and otherwise as the number of parts larger than
the number of ones minus the number of ones.

1
 Let p(n) denote the number of partitions of n. Dyson [16] first conjectured and later
confirmed by Atkin and Swinnerton-Dyer [7] that the rank can provide combinatorial
interpretations to the first two Ramanujan’s congruences p(5n + 4) ≡ 0 (mod 5) and
p(7n + 5) ≡ 0 (mod 7). Andrews and Garvan [4] showed that the crank may combina-
torially interpret the Ramanujan’s congruences p(5n + 4) ≡ 0 (mod 5), p(7n + 5) ≡ 0
(mod 7) and p(11n + 6) ≡ 0 (mod 11). Relations between ranks and cranks of partitions
have been studied by several authors, for example, Andrews and Lewis [5], Garvan [18],
Lewis [24, 25].
Let N(m, n) denote the number of partitions of n with rank m, and for n > 1, let
M(m, n) denote the number of partitions of n with crank m, and for n = 1, set

M(0, 1) = −1, M(−1, 1) = M(1, 1) = 1,

and M(m, 1) = 0 for m 6= 0, −1, 1.

Andrews and Garvan [4] derived the following generating function of M(m, n):
∞ ∞
X X (q; q)∞
M(m, n)z m q n = . (1.1)
m=−∞ n=0
(zq; q)∞ (q/z; q)∞

Here and throughout the rest of this paper, we adopt the common q-series notation [1]:
∞
Y (a; q)∞
(a; q)∞ = (1 − aq n ) and (a; q)n = .
n=0
(aq n ; q)∞

Chan and Mao [10] discovered the inequality on N(m, n) as stated below.

Theorem 1.1 (Chan and Mao). For any nonnegative integers m and n,

N(m, n) ≥ N(m + 2, n),

and for any nonnegative integers m and n such that n ≥ 12, n 6= m + 2,

N(m, n) ≥ N(m, n − 1).

Recently, Ji and Zang [23] proved the following unimodal property of the crank on
ordinary partitions.

Theorem 1.2 (Ji and Zang). For n ≥ 44 and 1 ≤ m ≤ n − 1, we have

M(m − 1, n) ≥ M(m, n). (1.2)

Ji and Zang also give the following monotonicity property of M(m, n).

2
Theorem 1.3 (Ji and Zang). For any n ≥ 14 and 0 ≤ m ≤ n − 2, we have

M(m, n) ≥ M(m, n − 1). (1.3)

Our interest in this paper is to consider an analogue of Theorem 1.2 and Theorem
1.3 for overpartitions. Specifically, we will investigate the unimodal property for the first
residue crank and the second residue crank. Recall that Corteel and Lovejoy [13] defined
an overpartition of n as a partition of n in which the first occurrence of a part may be
overlined. For example, there are 14 overpartitions of 4:
(4) (4) (3, 1) (3, 1) (3, 1) (3, 1) (2, 2)
(2, 2) (2, 1, 1) (2, 1, 1) (2, 1, 1) (2, 1, 1) (1, 1, 1, 1) (1, 1, 1, 1)

Analogous to the crank of an ordinary partition, Bringmann, Lovejoy and Osburn [9]
defined the first and second residual crank of an overpartition. The first residual crank
of an overpartition is defined as the crank of the subpartition consisting of non-overlined
parts. The second residual crank is defined as the crank of the subpartition consisting of
all of the even non-overlined parts divided by two.
For example, for λ = (9, 9, 7, 6, 5, 5, 4, 4, 3, 1, 1, 1), the partition consisting of non-
overlined parts of λ is (9, 7, 5, 5, 4, 3, 1, 1). The first residual crank of λ is 4. The partition
formed by even non-overlined parts of λ divided by two is (2). So the second residual
crank of λ is 2.
Since then, the researches on overpartitions have been extensively studied (see, for
example, Corteel [11], Corteel and Lovejoy [12, 13]). In addition, there have been a
great wealth of further results regarding the relations between the rank and crank of
overpartitions (see, for example, Andrews, Chan, Kim and Osburn [2], Jennings-Shaffer
[22], Lovejoy and Osburn [26, 27]), their connections to Maass form and mock theta
functions (see, for example, Andrews, Dixit, Schultz and Yee [3], Bringmann and Lovejoy
[8], Jennings-Shaffer [22]), and the relations between spt-function and overpartitions (see,
for example, Garvan and Jennings-Shaffer [19], Jennings-Shaffer [20, 21]).
Let M (m, n) (resp. M2(m, n)) denote the number of overpartitions of n with first
(resp. second) residual crank equal to m. Here we make the appropriate modifications
based on the fact that for ordinary partitions we have M(0, 1) = −1 and M(−1, 1) =
M(1, 1) = 1. For example, the overpartition (7, 5, 2, 1) contributes a −1 to the count of
M (0, 15) and a +1 to both M (−1, 15) and M (1, 15). For another example, (10, 9, 9, 7, 7,
6, 5, 3, 3, 2, 2) contributes a −1 to the count of M2(0, 57) and a +1 to both M2(−1, 57)
and M2(1, 57).
Bringmann, Lovejoy and Osburn [9] derived the following generating function of
M (m, n) and M2(m, n):
∞ ∞
X X (−q; q)∞ (q; q)∞
M (m, n)z m q n = . (1.4)
m=−∞ n=0
(zq; q)∞ (q/z; q)∞

3
 ∞ ∞
X X (−q; q)∞ (q; q)∞
M2(m, n)z m q n = . (1.5)
m=−∞ n=0
(q; q 2 )∞ (zq; q)∞ (q/z; q)∞

From (1.1), (1.4) and (1.5), for fixed m, the generating function of M (m, n) is equal
to
∞
X ∞
X
n
M (m, n)q = (−q; q)∞ M(m, n)q n (1.6)
n=0 n=0

and
∞ ∞
X (−q; q)∞ X
n
M2(m, n)q = 2)
M(m, n)q 2n . (1.7)
n=0
(q; q ∞ n=0

The main result of this paper is an analogue of Theorem 1.2 for overpartitions as
stated below.
Theorem 1.4. For any m ≥ 1 and n ≥ 0,
M(m − 1, n) ≥ M (m, n), (1.8)
except for (m, n) = (1, 1) or (m, n) = (1, 2).
Theorem 1.5. For any m ≥ 1 and n ≥ 0,
M2(m − 1, n) ≥ M2(m, n). (1.9)

The proofs of Theorem 1.4 and Theorem 1.5 are based on the following Lemma. For
the remainder of this paper, let {bn }∞n=0 and {cn }n=0 be any sequence of nonnegative
∞

integers but not necessarily the same sequence in different equations.

Lemma 1.6. The generating function of M(m − 1, n) − M(m, n) can be decomposed as
follows. For m = 1,
∞
X
(M(0, n) − M(1, n))q n = (1 − q)2 + q 2 (1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )
n=0
∞
X ∞
X
n
+ (1 − q) bn q + cn q n . (1.10)
n=0 n=0

For m = 2,
∞
X ∞
X ∞
X
n 3 n
(M(1, n) − M(2, n))q = q(1 − q)(1 − q ) + (1 − q) bn q + cn q n . (1.11)
n=0 n=0 n=0

For any m ≥ 3,
∞
X ∞
X ∞
X
n n
(M(m − 1, n) − M(m, n))q = (1 − q) bn q + cn q n . (1.12)
n=0 n=0 n=0

4
 In light of Lemma 1.6 and the relations (1.6) and (1.7), we give a proof of Theorem
1.4 and Theorem 1.5. Moreover, we also obtain the following monotonicity property for
M (m, n) and M2(m, n), which is an overpartition analogue of Theorem 1.3.
Theorem 1.7. For any m ≥ 0 and n ≥ 1,
M (m, n) ≥ M (m, n − 1) (1.13)
and
M2(m, n) ≥ M2(m, n − 1). (1.14)

As an application, we will use Theorem 1.4 to prove the unimodality of k-crank which
was conjectured by Fu and Tang [17]. Recall that the k-colored partition is a k-tuple of
partitions λ = (λ(1) , λ(2) , . . . , λ(k) ). For k ≥ 2, Fu and Tang [17] defined the k-crank of
k-colored partition as follows:
k-crank(λ) = ℓ(λ(1) ) − ℓ(λ(2) ), (1.15)
where ℓ π (i) denotes the number of parts in π (i) .

Let Mk (m, n) denote the number of k-colored partitions of n with k-crank m. The
generating function of Mk (m, n) can be derived by Bringmann and Dousse [14]:
∞ ∞
X X (q; q)2−k
Mk (m, n)z m q n = ∞
. (1.16)
n=0 m=−∞
(zq; q)∞ (z q; q)∞
−1

Fu and Tang [17, Conjecture 4.1] raised the following conjecture on {Mk (m, n)}nm=−n :
Conjecture 1.8 (Fu and Tang, 2018). For n ≥ 0 and k ≥ 2, the sequence of {Mk (m, n)}nm=−n
is unimodal except for n = 1, k = 2.

Fu and Tang also pointed out that using the asymptotic formula in [15] due to Bring-
mann and Manschot [15], one may give an asymptotic proof of Conjecture 1.8.
In this paper, we confirm Conjecture 1.8 with the aid of Theorem 1.4.
This paper is organized as follows: In Section 2, we give a proof of Lemma 1.6 with
the aid of Theorem 1.2. Section 3 is devoted to prove Theorem 1.4. In Section 4, we
demonstrate that Theorem 1.5 can be deduced by Theorem 1.4. In Section 5, we prove
Theorem 1.7. Finally, in Section 6, Conjecture 1.8 will be confirmed with the aid of
Theorem 1.4.

2 The transformation of the generating function of

M (m − 1, n) − M (m, n)
In this section, we give a proof of Lemma 1.6 in light of Theorem 1.2.

5
Proof of Lemma 1.6. From Theorem 1.2, it is routine to check that
∞
X
(M(0, n) − M(1, n))q n
n=0
= 1 − 2q + q 3 + q 4 − q 7 − q 9 + q 10 − q 11 + 2q 12 − q 13

+ 2q 14 − q 15 + 2q 16 − 2q 17 + 3q 18 − 3q 19 + 3q 20 − 2q 21

+ 3q 22 − 3q 23 + 6q 24 − 4q 25 + 6q 26 − 2q 27 + 7q 28 − 4q 29

+ 11q 30 − 5q 31 + 12q 32 − 3q 33 + 13q 34 − 4q 35 + 20q 36 − 6q 37

∞
X
+ 22q 38 − q 39 + 27q 40 − 3q 41 + 37q 42 − q 43 + bn q n . (2.1)
n=44

Set

f (q) = q 10 + q 14 + 2q 16 + 3q 18 + 2q 20 + 3q 22 + 4q 24 + 2q 26
+ 4q 28 + 5q 30 + 3q 32 + 4q 34 + 6q 36 + q 38 + 3q 40 + q 42

and

h(q) = q 14 + q 20 + 2q 24 + 4q 26 + 3q 28 + 6q 30
+ 9q 32 + 9q 34 + 14q 36 + 21q 38 + 24q 40 + 36q 42 .

It is trivial to check that

∞
X
(M(0, n) − M(1, n))q n = (1 − q)2 + q 2 (1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )
n=0

∞
X
+ (1 − q)f (q) + h(q) + bn q n . (2.2)
n=44

Clearly f (q) and h(q) have nonnegative coefficients. This yields (1.10).
We next assume that m = 2. Similar as above, by Theorem 1.2, it can be checked
that
∞
X
(M(1, n) − M(2, n))q n = q − q 2 − q 4 + q 5 + q 7 + q 9 − q 10 + q 11 − q 12 + q 13 − q 14
n=0

+ 2q 15 − q 16 + 3q 17 − q 18 + 4q 19 − q 20 + 5q 21 − q 22
∞
X
23 24 25 26
+ 6q −q + 8q −q + bn q n . (2.3)
n=27

6
Set
f (q) = q 9 + q 11 + q 13 + q 15 + q 17 + q 19 + q 21 + q 23 + q 25
and
h(q) = q 7 + q 15 + 2q 17 + 3q 19 + 4q 21 + 5q 23 + 7q 25 .
Together with (2.3), we see that
∞
X ∞
X
(M(1, n) − M(2, n))q n = q(1 − q)(1 − q 3 ) + (1 − q)f (q) + h(q) + bn q n .
n=0 n=27

This yields (1.11).

For 3 ≤ m ≤ 7, it is routine to check that (1.12) is valid. We next assume that
m ≥ 8. From Theorem 1.2, we see that for m ≥ 43 and n ≥ m + 1, we see that
M(m−1, n)−M(m, n) ≥ 0. It can be checked that when 8 ≤ m ≤ 42 and 44 ≥ n ≥ m+1,
M(m − 1, n) − M(m, n) ≥ 0 also holds. When n = m or m − 1, we have

M(m − 1, m − 1) − M(m, m − 1) = 1

and
M(m − 1, m) − M(m, m) = −1.
Note that for n ≤ m − 2, M(m − 1, n) = M(m, n) = 0. Thus
∞
X ∞
X
n m−1
(M(m − 1, n) − M(m, n)) q = q (1 − q) + bn q n . (2.4)
n=0 n=m+1

This completes the proof.

3 The proof of Theorem 1.4

This section is devoted to give a proof of Theorem 1.4. The main difficulty of the proof
is to deal with the case m = 1. To solve this difficulty, we need two more lemmas.
Lemma 3.1. The coefficients of q n in

(1 − q)2 (−q; q)∞

is nonnegative for all n ≥ 0 except for n = 1 or 4.

Proof. It is clear that

(1 − q)2 (−q; q)∞ = (1 − q)(−q; q)∞ − q(1 − q)(−q; q)∞ . (3.1)

We further expand the (1 − q)(−q; q)∞ and q(1 − q)(−q; q)∞ as follows.

7
 We first consider (1 − q)(−q; q)∞ . Let d(n) denote the number of distinct partitions
of n. Clearly
X∞
d(n)q n = (−q; q)∞ . (3.2)
n=0

Moreover, we may classify the set of distinct partitions based on the largest part. To be
specific, for j ≥ 1, let dj (n) denote the number of distinct partitions of n with the largest
part is equal to j. Then it is clear that
∞
X
dj (n)q n = q j (−q; q)j−1. (3.3)
n=0

Thus from (3.2) and (3.3) we see that

∞
X ∞ X
X ∞ ∞
X
n n
(−q; q)∞ = d(n)q = dj (n)q = 1 + q j (−q; q)j−1 . (3.4)
n=0 j=0 n=0 j=1

Thus we have

(1 − q)(−q; q)∞
∞
X ∞
X
=1 − q + q j (−q; q)j−1 − q j+1 (−q; q)j−1
j=1 j=1
∞
X ∞
X
j
=1 − q + q + q (−q; q)j−1 − q j (−q; q)j−2
j=2 j=2
∞
X
=1 + q 2j−1(−q; q)j−2
j=2
∞
X
3 5
=1 + q + q (1 + q) + q 2j−1 (1 + q)(1 + q 2 )(−q 3 ; q)j−4. (3.5)
j=4

Moreover,
∞
X
q 2j−1 (1 + q)(1 + q 2 )(−q 3 ; q)j−4
j=4
X∞ ∞
X ∞
X
2j 2 2j+1 3
= q (−q ; q)j−3 + q (−q ; q)j−4 + q 2j−1 (−q 3 ; q)j−4. (3.6)
j=4 j=4 j=4

Substituting (3.6) into (3.5), we derive that

(1 − q)(−q; q)∞

8
 ∞
X ∞
X ∞
X
=1 + q 3 + q 5 (1 + q) + q 2j (−q 2 ; q)j−3 + q 2j+1 (−q 3 ; q)j−4 + q 2j−1(−q 3 ; q)j−4.
j=4 j=4 j=4
(3.7)

On the other hand, from (3.5), we see that

∞
X
4 6
q(1 − q)(−q; q)∞ = q + q + q (1 + q) + q 2j (1 + q)(1 + q 2 )(−q 3 ; q)j−4. (3.8)
j=4

We deduce that
∞
X
q 2j (1 + q)(1 + q 2 )(−q 3 ; q)j−4
j=4
∞
X ∞
X ∞
X
2j 2 2j+1 3
= q (−q ; q)j−3 + q (−q ; q)j−4 + q 2j+3(−q 3 ; q)j−4. (3.9)
j=4 j=4 j=4

Substituting (3.9) into (3.8), we have

q(1 − q)(−q; q)∞

∞
X ∞
X ∞
X
4 6 2j 2 2j+1 3
=q + q + q (1 + q) + q (−q ; q)j−3 + q (−q ; q)j−4 + q 2j+3(−q 3 ; q)j−4.
j=4 j=4 j=4
(3.10)

Substituting (3.7) and (3.10) into (3.1), we deduce that

(1 − q)2 (−q; q)∞

∞
X ∞
X
3 4 5 7 2j−1 3
= 1−q+q −q +q −q + q (−q ; q)j−4 − q 2j+3 (−q 3 ; q)j−4
j=4 j=4
∞
X
= 1 − q + q 3 − q 4 + q 5 + q 9 + q 12 + q 2j−1 (−q 3 ; q)j−6 (q j−3 + q j−2 + q 2j−5 ). (3.11)
j=6

Clearly (3.11) has nonnegative coefficients in q n for all n ≥ 0 except for n = 1 or n = 4.

This completes the proof.

Lemma 3.2. The coefficient of q n in

(1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )(−q; q)∞

is nonnegative for all n ≥ 1.

9
Proof. From the Euler identity [1],
1
(−q; q)∞ = , (3.12)
(q; q 2 )∞

we deduce that

(1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )(−q; q)∞

(1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )
=
(q; q 2)∞
(−1 + q 2 + q 3 + q 4 − q 5 )
=
(1 − q 3 )(q 7 ; q 2 )∞
q4 (1 − q 2 )
= − . (3.13)
(1 − q 3 )(q 7 ; q 2)∞ (q 7 ; q 2 )∞

Let p7 (n) denote the number of partitions of n with each part odd and not less than 7.
It is clear that ∞
X 1
p7 (n)q n = 7 2 .
n=0
(q ; q )∞
Moreover, we may divide the set of partitions counted by p7 (n) into nonintersecting subsets
based on the largest part. To be specific, for odd j ≥ 7, let p7,j (n) denote the number of
partitions counted by p7 (n) such that the largest part is equal to j. It is easy to see that
∞
X qj
p7,j (n)q n = .
n=0
(q 7 ; q 2 )(j−5)/2

Thus we have
∞
1 XX
n
X qj
=1+ p7,j (n)q = 1 + . (3.14)
(q 7 ; q 2 )∞ j≥7 n=0 j≥7
(q 7 ; q 2 )(j−5)/2
j odd j odd

Using (3.14), we may transform the second term in (3.13) as follows.

1 − q2 X qj 2
X q j+2
= 1 + − q −
(q 7 ; q 2 )∞ j≥7
(q 7 ; q 2)(j−5)/2 j≥7
(q 7 ; q 2)(j−5)/2
j odd j odd

7 j
q X q X qj
= 1 − q2 + + −
1 − q7 j≥9
(q 7 ; q 2 )(j−5)/2 j≥9
(q 7 ; q 2 )(j−7)/2
j odd j odd

q7 qj
 
2
X 1
=1−q + + −1
1 − q7 j≥9
(q 7 ; q 2 )(j−7)/2 1 − qj
j odd

10
 q7 X q 2j
= 1 − q2 + +
1 − q7 j≥9
(q 7 ; q 2 )(j−5)/2
j odd

7
q2 q 18 X q 2j
=1−q + + + . (3.15)
1 − q 7 (1 − q 7 )(1 − q 9 ) j≥11 (q 7 ; q 2 )(j−5)/2
j odd

We next transform the first term in (3.13) in light of (3.14) as follows.

 
q4 q4  q7 q9 X qj
= 1 + + +

3 7 2
(1 − q )(q ; q )∞ 1−q 3 
1−q 7 7 9 7 2
(1 − q )(1 − q ) j≥11 (q ; q )(j−5)/2

j odd

4 11 13
q q q
= + +
1 − q 3 (1 − q 3 )(1 − q 7 ) (1 − q 3 )(1 − q 7 )(1 − q 9 )
X q j+4
+
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−5)/2
j odd

q 10 q6
    11
q
= q4 + q7 + + 1+q + 3
1 − q3 1−q 3 1 − q7
q 13 X q j+4
+ +
(1 − q 3 )(1 − q 7 )(1 − q 9 ) j≥11 (1 − q 3 )(q 7 ; q 2 )(j−5)/2
j odd

q 10 q 11 q 14 q 17
= q4 + q7 + + + +
1 − q 3 1 − q 7 1 − q 7 (1 − q 3 )(1 − q 7 )
q 13 X q j+4
+ + . (3.16)
(1 − q 3 )(1 − q 7 )(1 − q 9 ) j≥11 (1 − q 3 )(q 7 ; q 2 )(j−5)/2
j odd

Notice that
q 13 q 13 q 16
= + . (3.17)
(1 − q 3 )(1 − q 7 )(1 − q 9 ) (1 − q 7 )(1 − q 9 ) (1 − q 3 )(1 − q 7 )(1 − q 9 )
Moreover,
q j+4 q j+4 q j−4
X X  
= 1+
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−5)/2 j≥11
(1 − q 3 )(q 7 ; q 2 )(j−11)/2 (q j−2; q 2 )2 1 − q j−4
j odd j odd

X q j+4
=
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−11)/2 (q j−2; q 2 )2
j odd

X q 2j
+ . (3.18)
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−5)/2
j odd

11
Substituting (3.17) and (3.18) into (3.16), we see that

q4 4 q 10 q 11 q7 q 17
= q + + + +
(1 − q 3 )(q 7 ; q 2 )∞ 1 − q 3 1 − q 7 1 − q 7 (1 − q 3 )(1 − q 7 )
q 13 q 16
+ +
(1 − q 7 )(1 − q 9 ) (1 − q 3 )(1 − q 7 )(1 − q 9 )
X q j+4
+
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−11)/2 (q j−2 ; q 2 )2
j odd

X q 2j
+ . (3.19)
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−5)/2
j odd

By (3.15) and (3.19), we see that (3.13) can be rewritten as follows:

(1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )(−q; q)∞

q 10 q 17
= −1 + q 2 + q 4 + +
1 − q 3 (1 − q 3 )(1 − q 7 )
∞
X q j+4 q 16
+ +
j=11
(1 − q 3 )(q 7 ; q 2 )(j−11)/2 (q j−2; q 2 )2 (1 − q 3 )(1 − q 7 )(1 − q 9 )
j odd

q 11 q 13 q 18
 
+ + −
1 − q 7 (1 − q 7 )(1 − q 9 ) (1 − q 7 )(1 − q 9 )
 
∞ ∞
X q 2j X q 2j
+ − . (3.20)

3 7 2
(1 − q )(q ; q )(j−5)/2 7 2
(q ; q )(j−5)/2
j=11 j=11
j odd j odd

It is clear that
q 18 q 18 q9 q 18 q 27
 
= 1+ = + . (3.21)
(1 − q 7 )(1 − q 9 ) 1 − q7 1 − q9 1 − q 7 (1 − q 7 )(1 − q 9 )

Thus we have
q 11 q 13 q 18
+ −
1 − q 7 (1 − q 7 )(1 − q 9 ) (1 − q 7 )(1 − q 9 )
q 11 q 18 q 13 q 27
= − + −
1 − q 7 1 − q 7 (1 − q 7 )(1 − q 9 ) (1 − q 7 )(1 − q 9 )
q 13 (1 + q 7 )
= q 11 + . (3.22)
1 − q9

12
Moreover, it is clear that
X q 2j X q 2j X q 2j+3
− = . (3.23)
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−5)/2 j≥11
(q 7; q2)
(j−5)/2 j≥11
(1 − q 3 )(q 7 ; q 2 )
(j−5)/2
j odd j odd j odd

Substituting (3.22) and (3.23) into (3.20), we deduce that

(1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )(−q; q)∞

q 10 q 17 q 16
= −1 + q 2 + q 4 + + +
1 − q 3 (1 − q 3 )(1 − q 7 ) (1 − q 3 )(1 − q 7 )(1 − q 9 )
X q j+4 11 q 13 (1 + q 7 )
+ + q +
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−11)/2 (q j−2; q 2 )2 1 − q9
j odd

X q 2j+3
+ . (3.24)
j≥11
(1 − q 3 )(q 7 ; q 2 )(j−5)/2
j odd

Clearly, (3.24) has nonnegative coefficients in q n for all n ≥ 1. This completes the proof.

We are now in a position to prove Theorem 1.4.

Proof of Theorem 1.4. From (1.6), it is clear to see that for fixed integer m,
∞
X ∞
X
n
(M(m − 1, n) − M(m, n)) q n .


M (m − 1, n) − M (m, n) q = (−q; q)∞ (3.25)
n=0 n=0

From (3.25) and Lemma 1.6, we see that for m ≥ 3,

∞ ∞ ∞
!
X X X
(M (m − 1, n) − M (m, n))q n = (−q, q)∞ (1 − q) bn q n + cn q n . (3.26)
n=0 n=0 n=0

From (3.12), we deduce that

∞ n n
P P∞
X (1 − q) ∞ n n=0 bn q n=0 cn q
(M (m − 1, n) − M (m, n))q = +
n=0
(q; q 2)∞ (q; q 2 )∞
n n
P∞ P∞
n=0 bn q n=0 cn q
= + . (3.27)
(q 3 ; q 2 )∞ (q; q 2 )∞

Clearly, (3.27) has nonnegative coefficients. This yields

M (m − 1, n) ≥ M(m, n)

for m ≥ 3.

13
 Similar as above, for m = 2, using Lemma 1.6, (3.12) and (3.25), we see that
∞ ∞ ∞
!
X X X
(M(1, n) − M (2, n))q n = (−q, q)∞ q(1 − q)(1 − q 3 ) + (1 − q) bn q n + cn q n
n=0 n=0 n=0
n n
P∞ P∞
q n=0 bn q n=0 cn q
= + + . (3.28)
(q 5 ; q 2 )∞ (q 3 ; q 2 )∞ (q; q 2 )∞
Clearly (3.28) has nonnegative coefficients. This yields Theorem 1.4 holds for m = 2.
Finally, for m = 1, by Lemma 1.6, (3.12) and (3.25), we deduce that
∞
X
(M (0, n) − M (1, n))q n
n=0
= q 2 (−q; q)∞ (1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )
∞ ∞
1−q X X
+ (1 − q)2 (−q; q)∞ + bn q n
+ (−q; q) ∞ cn q n
(q; q 2 )∞ n=0 n=0

= q 2 (−q; q)∞ (1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 )

∞ ∞
2 1 X
n
X
+ (1 − q) (−q; q)∞ + 3 2 bn q + (−q; q)∞ cn q n . (3.29)
(q ; q )∞ n=0 n=0

From Lemma 3.1 and Lemma 3.2 we see that

q 2 (−q; q)∞ (1 − q)(1 − q 5 )(−1 + q 2 + q 3 + q 4 − q 5 ) + (1 − q)2 (−q; q)∞
has nonnegative coefficients in q n except for n = 1, 2, 4. Thus for n 6= 1, 2, 4, the coefficient
of q n in (3.29) is nonnegative. It is trivial to check that M(0, 4) = M (1, 4) = 2. This
completes the proof.

4 The proof of Theorem 1.5

In this section, we give a proof of Theorem 1.5 with the aid of Theorem 1.4.
Proof of Theorem 1.5. By (1.7), it is clear that for any fixed integer m,
∞
X
M2(m − 1, n) − M2(m, n) q n


n=0
∞
(−q; q)∞ X
= (M(m − 1, n) − M(m, n)) q 2n
(q; q 2 )∞ n=0
∞
!
(−q; q 2 )∞ X
= (−q 2 ; q 2 )∞ (M(m − 1, n) − M(m, n))q 2n . (4.1)
(q; q 2)∞ n=0

14
From (3.25), we see that
∞
X
(M2(m − 1, n) − M2(m, n))q n
n=0
∞
!
2
(−q; q )∞ X
M(m − 1, n) − M (m, n) q 2n


= . (4.2)
(q; q 2 )∞ n=0

Hence by Theorem 1.4, for m ≥ 2,

∞
X
(M2(m − 1, n) − M2(m, n))q n
n=0

has nonnegative coefficients. This yields Theorem 1.5 holds for m ≥ 2.

We next consider the case m = 1. In this case, by Theorem 1.4, it is easy to check
that ∞ ∞
X
2n X
2 4 6
M (m − 1, n) − M (m, n) q = 1 − q − q + q + bn q n . (4.3)
n=0 n=8
n
Thus it suffices to show that the coefficients of q in
(−q; q 2 )∞
(1 − q 2 − q 4 + q 6 )
(q; q 2)∞
is nonnegative for all n ≥ 0. Since
(−q; q 2 )∞ 2 4 6 (−q; q 2 )∞
(1 − q − q + q ) = (1 − q 2 )(1 − q 4 )
(q; q 2 )∞ (q; q 2 )∞
1+q
= (−q; q 2 )∞ (1 − q 4 ) · 3 2 . (4.4)
(q ; q )∞
We claim that
(−q; q 2 )∞ (1 − q 4 )
has nonnegative coefficients in q n for all n ≥ 0.
Let sc(n) denote the number of distinct odd partitions of n. It is clear that
∞
X
2
(−q; q )∞ = sc(n)q n .
n=0

We next divide the set of partitions counted by sc(n) into nonintersecting subsets based
on the largest part. For j ≥ 1 odd, let scj (n) denote the number of distinct odd partitions
of n with largest part equal to j. Then
∞
X
scj (n)q n = q j (−q; q 2 )(j−1)/2 . (4.5)
n=0

15
So we see that X
(−q; q 2 )∞ = 1 + q j (−q; q 2 )(j−1)/2 . (4.6)
j≥1
j odd

Hence we deduce that

X X
(1 − q 4 )(−q; q 2 ) = 1 − q 4 + q j (−q; q 2 )(j−1)/2 − q j+4(−q; q 2 )(j−1)/2
j≥1 j≥1
j odd j odd
X X
= 1 − q 4 + q + q 3 (1 + q) + q j (−q; q 2 )(j−1)/2 − q j (−q; q 2 )(j−5)/2
j≥5 j≥5
j odd j odd
X
= 1 + q + q3 + q j (−q; q 2 )(j−5)/2 (q j−4 + q j−2 + q 2j−6). (4.7)
j≥5
j odd

Clearly, (4.7) has nonnegative coefficients, and our claim follows.

From the above claim, it is clear to see that (4.4) has nonnegative coefficients of q n
for all n ≥ 0. This completes the proof.

5 The monotonicity properties of M(m, n) and M 2(m, n)

In this section, we give a proof of Theorem 1.7.
Proof of Theorem 1.7. For fixed integer m,
X∞ ∞
X
n
(M (m, n + 1) − M (m, n))q = (1 − q) M(m, n)q n . (5.1)
n=0 n=0

From (1.6),
∞
X ∞
X
n
(M (m, n) − M (m, n − 1))q = (1 − q)(−q; q)∞ M(m, n)q n . (5.2)
n=1 n=0

Thus by (3.12), we have

∞ ∞
X
n (1 − q) X
(M (m, n) − M(m, n − 1))q = 2
M(m, n)q n
n=1
(q; q )∞ n=0
∞
1 X
= 3 2 M(m, n)q n , (5.3)
(q ; q )∞ n=0
which clearly has nonnegative coefficients. This yields (1.13). Using the same argument,
we deduce that
∞ ∞
X
n (−q; q)∞ X
(M2(m, n) − M2(m, n − 1))q = 3 2 M(m, n)q 2n , (5.4)
n=1
(q ; q )∞ n=0
which implies (1.14). This completes the proof.

16
6 The proof of Conjecture 1.8
In this section, we prove Conjecture 1.8 with the aid of Theorem 1.4.
Proof of Conjecture 1.8. From (1.4) and (1.16), it is clear that when k ≥ 2
∞ ∞
X X (q; q)2−k
Mk (m, n)z m q n = ∞

m=−∞ n=0
(zq; q)∞ (q/z; q)∞
(−q; q)∞ (q; q)∞ 1
= · 2 2 k−2
(zq; q)∞ (q/z; q)∞ (q ; q )∞ (q; q)∞
∞ ∞
1 X X
= 2 2 k−2
M(m, n)z m q n . (6.1)
(q ; q )∞ (q; q)∞ m=−∞ n=0

Equating the coefficient of z m on both sides, we find that for fixed m,

∞ ∞
X
n 1 X
Mk (m, n)q = 2 2 k−2
M(m, n)q n . (6.2)
n=0
(q ; q ) ∞ (q; q) ∞ n=0

Thus
∞ ∞
X
n1 X
(Mk (m − 1, n) − Mk (m, n))q = 2 2 k−2
(M (m − 1, n) − M (m, n))q n .
n=0
(q ; q )∞ (q; q)∞ n=0
(6.3)
When m ≥ 2, by Theorem 1.4, we find that M(m − 1, n) − M (m, n) ≥ 0. Thus by
(6.3), we see that Mk (m − 1, n) ≥ Mk (m, n), as desired.
We now assume that m = 1. From Theorem 1.4 and simple calculation, we have
∞
X ∞
X
(M (0, n) − M (1, n))q n = 1 − q − q 2 + q 3 + q 5 + bn q n . (6.4)
n=0 n=6

Substituting (6.4) into (6.3),

∞
X
(Mk (0, n) − Mk (1, n))q n
n=0
∞
1 − q − q2 + q3 + q5 1 X
= + bn q n
(q 2 ; q 2 )∞ (q; q)∞
k−2 (q 2 ; q 2 )∞ (q; q)∞
k−2
n=6
∞
−q + q 3 + q 5
 
1 1 1 X
= + + bn q n . (6.5)
(q 4 ; q 2 )∞ (q 2 ; q 2 )∞ k−2
(q; q)∞ (q 2 ; q 2 )∞ (q; q)∞
k−2
n=6

We claim that for n ≥ 2, the coefficients of q n in

−q + q 3 + q 5
(q 2 ; q 2 )∞

17
is nonnegative. Recall that Andrews and Merca [6] proved the following inequality holds
for n > 0:
p(n) − p(n − 1) − p(n − 2) + p(n − 5) ≤ 0. (6.6)
Thus for n > 0,
p(n) − p(n − 1) − p(n − 2) ≤ 0. (6.7)
Therefore
∞ ∞
X
n−1 + q + q 2 X
(p(n − 1) + p(n − 2) − p(n))q = = −1 + bn q n . (6.8)
n=0
(q; q)∞ n=1

Setting q = q 2 in (6.8), we derive that

∞
−1 + q 2 + q 4 X
= −1 + bn q 2n . (6.9)
(q 2 ; q 2 )∞ n=1

Thus ∞
−q + q 3 + q 5 X
= −q + bn q 2n+1 . (6.10)
(q 2 ; q 2 )∞ n=1

This yields our claim.

Substituting (6.10) into (6.5), and notice that the constant term in 1/(q 4; q 2 )∞ is equal
to 1,
∞ ∞
X 1−q X
(Mk (0, n) − Mk (1, n))q n = k−2
+ bn q n . (6.11)
n=0
(q; q) ∞ n=0

Thus when k = 2, the coefficients of q n in (6.11) is nonnegative for n ≥ 2. When k ≥ 3,

∞ ∞
X 1 X
(Mk (0, n) − Mk (1, n))q n = k−3 (q 2 ; q)
+ bn q n , (6.12)
n=0
(q; q)∞ ∞ n=0

which has nonnegative coefficients. This completes the proof.

Acknowledgments. This work was supported by the National Science Foundation of

China (Grant NO. 11801119).

References
[1] G.E. Andrews, The Theory of Partitions, Addison-Wesley, 1976.
[2] G.E. Andrews, S.H. Chan, B. Kim and R. Osburn, The first positive rank and crank
moments for overpartitions, Ann. Combin. 20 (2) (2016) 193–207.
[3] G.E. Andrews, A. Dixit, D. Schultz and A.J. Yee, Overpartitions related to the mock theta
function ω(q), arXiv: 1603.04352.

18
 [4] G.E. Andrews and F.G. Garvan, Dyson’s crank of a partition, Bull. Amer. Math. Soc. 18
(2) (1988) 167–171.
[5] G.E. Andrews and R. Lewis, The ranks and cranks of partitions moduli 2, 3 and 4, J.
Number Theory 85 (1) (2000) 74–84.
[6] G.E. Andrews and M. Merca, The truncated pentagonal number theorem, J. Combin.
Theory Ser. A 119 (8) (2012) 1639–1643.
[7] A.O.L. Atkin and P. Swinnerton-Dyer, Some properties of partitions, Proc. Lond. Math.
Soc. 66 (1954) 84–106.
[8] K. Bringmann and J. Lovejoy, Dyson’s rank, overpartitions, and weak Maass forms, Int.
Math. Res. Not. (19) (2007) Art. ID rnm063 34 pp.
[9] K. Bringmann, J. Lovejoy and R. Osburn, Rank and crank moments for overpartitions, J.
Number Theory 129 (7) (2009) 1758–1772.
[10] S.H. Chan and R. Mao, Inequalities for ranks of partitions and the first moment of ranks
and cranks of partitions, Adv. Math. 258 (2014) 414–437.
[11] S. Corteel, Particle seas and basic hypergeometric series, Adv. in Appl. Math. 31 (1) (2003)
199–214.
[12] S. Corteel and J. Lovejoy, Frobenius partitions and the combinatorics of Ramanujan’s 1 ψ 1
summation, J. Combin. Theory Ser. A 97 (2002) 179–183.
[13] S. Corteel and J. Lovejoy, Overpartitions, Trans. Amer. Math. Soc. 356 (2004) 1623–1635.
[14] K. Bringmann and J. Dousse, On Dyson’s crank conjecture and the uniform asymptotic
behavior of certain inverse theta functions, Trans. Amer. Math. Soc. 368 (5) (2016) 3141–
3155.
[15] K. Bringmann and J. Manschot, Asymptotic formulas for coefficients of inverse theta func-
tions, Commun. Number Theory Phys. 7 (3) (2013) 497–513.
[16] F.J. Dyson, Some guesses in the theory of partitions, Eureka (Cambridge) vol. 8 (1944)
10–15.
[17] S. Fu and D. Tang, On a generalized crank for k-colored partitions, J. Number Theory 184
(2018) 485–497.
[18] F.G. Garvan, The crank of partitions mod 8, 9 and 10, Trans. Amer. Math. Soc. 322 (1)
(1990) 79–94.
[19] F.G. Garvan and C. Jennings-Shaffer, The spt-crank for overpartitions, Acta Arith. 166
(2) (2014) 141–188.
[20] C. Jennings-Shaffer, Another SPT crank for the number of smallest parts in overpartitions
with even smallest part, J. Number Theory 148 (2015) 196–203.
[21] C. Jennings-Shaffer, Higher order SPT functions for overpartitions, overpartitions with
smallest part even, and partitions with smallest part even and without repeated odd parts,
J. Number Theory 149 (2015) 285–312.
[22] C. Jennings-Shaffer, Overpartition rank differences modulo 7 by Maass forms, J. Number
Theory 163 (2016) 331–358.
[23] K.Q. Ji and W.J.T. Zang, Unimodality of the Andrews-Garvan-Dyson cranks of partitions,
submitted, arXiv:1811.07321.
[24] R. Lewis, On the rank and the crank modulo 4, Proc. Amer. Math. Soc. 112 (4) (1991)
925–933.

19
[25] R. Lewis, On some relations between the rank and the crank, J. Combin. Theory Ser. A
59 (1) (1992) 104–110.
[26] J. Lovejoy and R. Osburn, Rank differences for overpartitions, Quart. J. Math. 59 (2008)
257–273.
[27] J. Lovejoy and R. Osburn, M2 -rank differences for overpartitions, Acta Arith. 144 (2)
(2010) 193–212.

20